论文部分内容阅读
为了对矿井突水水源进行准确、高效的判别,综合考虑水化学特征,选取Ca~(2+),Mg~(2+),K~++Na~+,HCO-3,SO2-4,Cl~-和总硬度7个指标的质量浓度(mg/L)作为矿井突水水源的最初判别指标。利用粗糙集(RS)理论的属性约简来筛选水化学特征指标,用以作为水源识别的核心判别指标,建立基于RS的矿井突水水源识别的最小二乘支持向量机(LSSVM)模型。选用约简处理后的13组煤矿数据对模型进行训练,再用训练好的模型对另外12组突水数据进行水源判别,并与未进行属性约简的LSSVM模型及Fisher判别分析法、随机森林方法进行对比。结果表明,利用属性约简方法可以很好地排除原始数据中的冗余信息干扰,因而能有效判别矿井突水水源,使矿井突水水源模型的误判率降低至0;而且指标约简过程可以降低LSSVM运算的复杂度,也能够提高判别效率。
In order to make accurate and efficient discrimination of mine water inrush, considering chemical characteristics of water, Ca2 +, Mg2 +, K ++ Na +, HCO-3, SO2-4, Cl ~ - and total hardness of seven indicators of the mass concentration (mg / L) as a preliminary identification of water inrush indicators. The attributes of water chemistry were screened by attribute reduction based on rough sets (RS) theory, which was used as the core discriminant index of water source identification. Based on RS, a least square support vector machine (LSSVM) model of mine water inrush source identification was established. Thirteen sets of coal mine data after reduction were used to train the model. Then another 12 sets of water inrush data were discriminated by trained model, and compared with the LSSVM model and Fisher discriminant analysis method without attribute reduction, Method to compare. The results show that the method of attribute reduction can effectively eliminate the redundant information interference in the original data and thus can effectively distinguish the water inrush from the mine and reduce the misjudgment rate of the mine water inrush model to zero. Moreover, the index reduction process Can reduce the complexity of LSSVM operation, but also can improve the discrimination efficiency.